Elamur Research⌕
☰
home / papers / 2025 / Hallucination Detection in LLMs Using Spectral Features of Attention Maps

Hallucination Detection in LLMs Using Spectral Features of Attention Maps

Paper · 2025 · Jakub Binkowski, Denis Janiak, Albert Sawczyn et al.

Summary

Large language models often produce wrong or unsupported answers, so reliable ways to flag these hallucinations are needed. Binkowski et al. treat each attention map of an LLM as a weighted directed graph over tokens and study its spectral properties. For every head and layer, they build a Laplacian using an out-degree matrix normalized by sequence position. Because this matrix is lower triangular, its eigenvalues are simply its diagonal entries. Their method, LapEigvals, takes the k largest eigenvalues from all heads and layers, concatenates them, compresses them with PCA and passes them to a logistic regression probe. Training labels come from QA datasets: an LLM judge (gpt-4o-mini), or exact match where possible, decides whether a generated answer is incorrect. Across seven QA benchmarks and five open models, the approach beats earlier detectors that rely only on attention, such as AttentionScore. It also stays stable across different values of k, prompts and sampling temperatures. Unlike sampling-based uncertainty methods, it needs just one generation pass.

Key points

  • Statistical motivation: On Llama-3.1-8B with TriviaQA, a two-sided Mann-Whitney U test showed significant differences (p < 0.05) between hallucinated and correct answers in 91% of heads for the top-10 Laplacian eigenvalues, versus 80% for AttentionScore. The same pattern held across 7 datasets and 5 LLMs.
  • Datasets: NQ-Open (3,610), TriviaQA (7,983), CoQA (5,928), SQuADv2 rc.nocontext (9,960), HaluEvalQA (10,000), TruthfulQA (817) and GSM8k test (1,319, scored by exact match).
  • Models and labeling: The models were Llama-3.1-8B, Llama-3.2-3B, Phi-3.5, Mistral-Nemo and Mistral-Small-24B, decoded at temperatures 0.1 and 1.0. Judge labels from gpt-4o-mini were cross-checked against gpt-4.1.
  • Limitations: The method requires labeled training data and white-box access to the model. Probes are tied to one layer/head configuration, inputs must have at least k tokens, and testing covered only English. Transfer across datasets requiring different skills is limited.
  • Code: The full implementation is at github.com/graphml-lab-pwr/lapeigvals.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓University of Technology SydneySydney · Australia18 works · 12,833 citations🎓Wrocław University of Science and TechnologyWroclaw · Poland8 works · 945 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

Cite

Cite as Jakub Binkowski, Denis Janiak, Albert Sawczyn, Bogdan Gabrys and Tomasz Jan Kajdanowicz. 2025. Hallucination Detection in LLMs Using Spectral Features of Attention Maps. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2025.emnlp-main.1239

@inproceedings{binkowski2025hallucination,
  title = {{Hallucination Detection in LLMs Using Spectral Features of Attention Maps}},
  author = {Jakub Binkowski, Denis Janiak, Albert Sawczyn, Bogdan Gabrys and Tomasz Jan Kajdanowicz},
  booktitle = {Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing},
  year = {2025},
  doi = {10.18653/v1/2025.emnlp-main.1239},
  url = {https://doi.org/10.18653/v1/2025.emnlp-main.1239}
}

References

55 references, 23 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Marah Abdin, Jyoti Aneja, Hany Awadalla, Ahmed Awadallah, Ammar Ahmad Awan, Nguyen Bach, Amit Bahree, Arash Bakhtiari, Jianmin Bao, Harkirat Behl, Alon Benhaim, Misha Bilenko, Johan Bjorck, Sébastien Bubeck, Martin Cai, Qin Cai, Vishrav Chaudhary, Dong Chen, Dongdong Chen, Weizhu Chen, Yen-Chun Chen… — Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone
  2. Guillaume Alain and Yoshua Bengio. 2016. Understanding intermediate layers using linear classifier probes. find
  3. Jason Ansel, Edward Yang, Horace He, Natalia Gimelshein, Animesh Jain, Michael Voznesensky, Bin Bao, Peter Bell, David Berard, Evgeni Burovski, Geeta Chauhan, Anjali Chourdia, Will Constable, Alban Desmaison, Zachary DeVito, Elias Ellison, Will Feng, Jiong Gong, Michael Gschwind, Brian Hirsh, Sherlo… find
  4. Gabriel Y. Arteaga, Thomas B. Schön, and Nicolas Pielawski. 2024. Hallucination Detection in LLMs: Fast and Memory-Efficient Finetuned Models. In Northern Lights Deep Learning Conference 2025. find
  5. Amos Azaria and Tom Mitchell. 2023. The Internal State of an LLM Knows When It‘s Lying. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 967–976, Singapore. Association for Computational Linguistics.
  6. Randall Balestriero, Mark Ibrahim, Vlad Sobal, Ari Morcos, Shashank Shekhar, Tom Goldstein, Florian Bordes, Adrien Bardes, Gregoire Mialon, Yuandong Tian, Avi Schwarzschild, Andrew Gordon Wilson, Jonas Geiping, Quentin Garrido, Pierre Fernandez, Amir Bar, Hamed Pirsiavash, Yann LeCun, and Micah Gold… — [arXiv:2304.12210](https://arxiv.org/abs/2304.12210) find
  7. Federico Barbero, Andrea Banino, Steven Kapturowski, Dharshan Kumaran, João G. M. Araújo, Alex Vitvitskyi, Razvan Pascanu, and Petar Veličković. 2024. Transformers need glasses! Information over-squashing in language tasks. arXiv preprint. ArXiv:2406.04267 [cs]. — [arXiv:2406.04267](https://arxiv.org/abs/2406.04267) find
  8. Mitchell Black, Zhengchao Wan, Amir Nayyeri, and Yusu Wang. 2023. Understanding Oversquashing in GNNs through the Lens of Effective Resistance. In International Conference on Machine Learning, pages 2528–2547. PMLR. ArXiv:2302.06835 [cs]. — [arXiv:2302.06835](https://arxiv.org/abs/2302.06835) find
  9. Joan Bruna, Wojciech Zaremba, Arthur Szlam, and Yann LeCun. 2013. Spectral Networks and Locally Connected Networks on Graphs. CoRR. find
  10. Chao Chen, Kai Liu, Ze Chen, Yi Gu, Yue Wu, Mingyuan Tao, Zhihang Fu, and Jieping Ye. 2024. find
  11. INSIDE: LLMs’ Internal States Retain the Power of Hallucination Detection. In The Twelfth International Conference on Learning Representations. find
  12. Yung-Sung Chuang, Linlu Qiu, Cheng-Yu Hsieh, Ranjay Krishna, Yoon Kim, and James R. Glass. 2024a. Lookback Lens: Detecting and Mitigating Contextual Hallucinations in Large Language Models Using Only Attention Maps. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Proce…
  13. Yung-Sung Chuang, Yujia Xie, Hongyin Luo, Yoon Kim, James R. Glass, and Pengcheng He. 2024b. DoLa: Decoding by Contrasting Layers Improves Factuality in Large Language Models. In The Twelfth International Conference on Learning Representations.
  14. Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, Christopher Hesse, and John Schulman. 2021. Training verifiers to solve math word problems. arXiv preprint arXiv:2110.14168.
  15. Tri Dao, Daniel Y. Fu, Stefano Ermon, Atri Rudra, and Christopher Ré. 2022. FLASHATTENTION: fast and memory-efficient exact attention with IO-awareness. In Proceedings of the 36th international conference on neural information processing systems, Nips ’22, New Orleans, LA, USA. Curran Associates Inc… find
  16. Sebastian Farquhar, Jannik Kossen, Lorenz Kuhn, and Yarin Gal. 2024. Detecting hallucinations in large language models using semantic entropy. Nature, 630(8017):625–630. Publisher: Nature Publishing Group. find
  17. Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, Amy Yang, Angela Fan, Anirudh Goyal, Anthony Hartshorn, Aobo Yang, Archi Mitra, Archie Sravankumar, Artem Korenev, Arthur Hinsvark, Arun Rao,… — The Llama 3 Herd of Models
  18. Lei Huang, Weijiang Yu, Weitao Ma, Weihong Zhong, Zhangyin Feng, Haotian Wang, Qianglong Chen, Weihua Peng, Xiaocheng Feng, Bing Qin, and Ting Liu. 2023. A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions. arXiv preprint. ArXiv:2311.05232 [cs].
  19. Ian T. Jolliffe and Jorge Cadima. 2016. Principal component analysis: a review and recent developments. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 374(2065):20150202. Publisher: Royal Society. find
  20. Mandar Joshi, Eunsol Choi, Daniel Weld, and Luke Zettlemoyer. 2017. TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1601–1611, Vancouver…
  21. Hazel Kim, Adel Bibi, Philip Torr, and Yarin Gal. 2024. Detecting LLM Hallucination Through Layer-wise Information Deficiency: Analysis of Unanswerable Questions and Ambiguous Prompts. arXiv preprint. ArXiv:2412.10246 [cs]. — [arXiv:2412.10246](https://arxiv.org/abs/2412.10246) find
  22. Jannik Kossen, Jiatong Han, Muhammed Razzak, Lisa Schut, Shreshth Malik, and Yarin Gal. 2024. Semantic Entropy Probes: Robust and Cheap Hallucination Detection in LLMs. arXiv preprint. ArXiv:2406.15927 [cs]. — [arXiv:2406.15927](https://arxiv.org/abs/2406.15927) find
  23. Ruslan Kuprieiev, skshetry, Peter Rowland, Dmitry Petrov, Pawel Redzynski, Casper da Costa-Luis, David de la Iglesia Castro, Alexander Schepanovski, Ivan Shcheklein, Gao, Batuhan Taskaya, Jorge Or-pinel, Fábio Santos, Daniele, Ronan Lamy, Aman Sharma, Zhanibek Kaimuldenov, Dani Hodovic, Nikita Koden… find
  24. Tom Kwiatkowski, Jennimaria Palomaki, Olivia Redfield, Michael Collins, Ankur Parikh, Chris Alberti, Danielle Epstein, Illia Polosukhin, Jacob Devlin, Kenton Lee, Kristina Toutanova, Llion Jones, Matthew Kelcey, Ming-Wei Chang, Andrew M. Dai, Jakob Uszkoreit, Quoc Le, and Slav Petrov. 2019. Natural…
  25. Minhyeok Lee. 2023. A Mathematical Investigation of Hallucination and Creativity in GPT Models. Mathematics, 11(10):2320. find
  26. Junyi Li, Jie Chen, Ruiyang Ren, Xiaoxue Cheng, Xin Zhao, Jian-Yun Nie, and Ji-Rong Wen. 2024. The Dawn After the Dark: An Empirical Study on Factuality Hallucination in Large Language Models. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long…
  27. Junyi Li, Xiaoxue Cheng, Wayne Xin Zhao, Jian-Yun Nie, and Ji-Rong Wen. 2023. HaluEval: A Large-Scale Hallucination Evaluation Benchmark for Large Language Models. arXiv preprint. ArXiv:2305.11747 [cs].
  28. Xun Liang, Shichao Song, Zifan Zheng, Hanyu Wang, Qingchen Yu, Xunkai Li, Rong-Hua Li, Feiyu Xiong, and Zhiyu Li. 2024. Internal Consistency and Self-Feedback in Large Language Models: A Survey. CoRR, abs/2407.14507. — [arXiv:2407.14507](https://arxiv.org/abs/2407.14507)
  29. Stephanie Lin, Jacob Hilton, and Owain Evans. 2022. TruthfulQA: Measuring How Models Mimic Human Falsehoods. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 3214–3252, Dublin, Ireland. Association for Computational Linguistics…
  30. Potsawee Manakul, Adian Liusie, and Mark Gales. 2023. SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models. In Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing, pages 9004–9017, Singapore. Association for Computation… find
  31. Henry B Mann and Donald R Whitney. 1947. On a test of whether one of two random variables is stochastically larger than the other. The annals of mathematical statistics, pages 50–60. Publisher: JSTOR. find
  32. Samuel Marks and Max Tegmark. 2024. The Geometry of Truth: Emergent Linear Structure in Large Language Model Representations of True/False Datasets. In First Conference on Language Modeling.
  33. Mistral AI Team. 2025. Mistral-small-24B-instruct-2501. find
  34. Mishkin, Vinnie Monaco, Evan Morikawa, Daniel Mossing, Tong Mu, Mira Murati, Oleg Murk, David Mély, Ashvin Nair, Reiichiro Nakano, Rajeev Nayak, Arvind Neelakantan, Richard Ngo, Hyeonwoo Noh, Long Ouyang, Cullen O’Keefe, Jakub Pachocki, Alex Paino, Joe Palermo, Ashley Pantuliano, Giambattista Parasc… — GPT-4 Technical Report
  35. Mistral AI Team and NVIDIA. 2024. Mistral-nemo-instruct-2407. find
  36. Kushan Mitra, Dan Zhang, Sajjadur Rahman, and Estevam Hruschka. 2024. FactLens: Benchmarking Fine-Grained Fact Verification. arXiv preprint. ArXiv:2411.05980 [cs]. find
  37. Bojan Mohar. 1997. Some applications of Laplace eigenvalues of graphs. In Geňa Hahn and Gert Sabidussi, editors, Graph Symmetry, pages 225–275. Springer Netherlands, Dordrecht. find
  38. OpenAI, Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, Red Avila, Igor Babuschkin, Suchir Balaji, Valerie Balcom, Paul Baltescu, Haiming Bao, Mohammad Bavarian, Jeff Belgum, Irwan Bello, Ja… find
  39. F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, and E. Duchesnay. 2011. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research, 12:282… find
  40. Pranav Rajpurkar, Robin Jia, and Percy Liang. 2018. Know What You Don‘t Know: Unanswerable Questions for SQuAD. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 784–789, Melbourne, Australia. Association for Computational Ling…
  41. Siva Reddy, Danqi Chen, and Christopher D. Manning. 2019. CoQA: A Conversational Question Answering Challenge. Transactions of the Association for Computational Linguistics, 7:249–266. Place: Cambridge, MA Publisher: MIT Press.
  42. Methods in Natural Language Processing: System Demonstrations, pages 38–45, Online. Association for Computational Linguistics. find
  43. Matthew Renze. 2024. The Effect of Sampling Temperature on Problem Solving in Large Language Models. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 7346–7356, Miami, Florida, USA. Association for Computational Linguistics.
  44. Xinyi Wu, Amir Ajorlou, Yifei Wang, Stefanie Jegelka, and Ali Jadbabaie. 2024. On the role of attention masks and LayerNorm in transformers. In Advances in neural information processing systems, volume 37, pages 14774–14809. Curran Associates, Inc. find
  45. Gaurang Sriramanan, Siddhant Bharti, Vinu Sankar Sadasivan, Shoumik Saha, Priyatham Kattakinda, and Soheil Feizi. 2024. LLM-Check: Investigating Detection of Hallucinations in Large Language Models. In The Thirty-eighth Annual Conference on Neural Information Processing Systems. find
  46. The pandas development team. 2020. dev/pandas: Pandas. pandas-Jake Topping, Francesco Di Giovanni, Benjamin Paul Chamberlain, Xiaowen Dong, and Michael M. Bronstein. 2022. Understanding over-squashing and bottlenecks on graphs via curvature. In International Conference on Learning Representations. find
  47. A Vaswani. 2017. Attention is all you need. Advances in Neural Information Processing Systems.
  48. Petar Veličković, Christos Perivolaropoulos, Federico Barbero, and Razvan Pascanu. 2024. softmax is not enough (for sharp out-of-distribution). arXiv preprint. ArXiv:2410.01104 [cs]. — [arXiv:2410.01104](https://arxiv.org/abs/2410.01104) find
  49. Ziwei Xu, Sanjay Jain, and Mohan Kankanhalli. 2024. Hallucination is Inevitable: An Innate Limitation of Large Language Models. arXiv preprint. ArXiv:2401.11817.
  50. Lianmin Zheng, Wei-Lin Chiang, Ying Sheng, Siyuan Zhuang, Zhanghao Wu, Yonghao Zhuang, Zi Lin, Zhuohan Li, Dacheng Li, Eric P. Xing, Hao Zhang, Joseph E. Gonzalez, and Ion Stoica. 2023. Judging LLM-as-a-judge with MT-bench and Chatbot Arena. In Proceedings of the 37th International Conference on Neu…
  51. Derui Zhu, Dingfan Chen, Qing Li, Zongxiong Chen, Lei Ma, Jens Grossklags, and Mario Fritz. 2024. PoLLMgraph: Unraveling Hallucinations in Large Language Models via State Transition Dynamics. In Findings of the Association for Computational Linguistics: NAACL 2024, pages 4737–4751, Mexico City, Mexi… find
  52. Pauli Virtanen, Ralf Gommers, Travis E. Oliphant, Matt Haberland, Tyler Reddy, David Cournapeau, Evgeni Burovski, Pearu Peterson, Warren Weckesser, Jonathan Bright, Stéfan J. van der Walt, Matthew Brett, Joshua Wilson, K. Jarrod Millman, Nikolay Mayorov, Andrew R. J. Nelson, Eric Jones, Robert Kern,… — SciPy 1.0 — Fundamental Algorithms for Scientific Computing in Python
  53. Ulrike von Luxburg. 2007. A tutorial on spectral clustering. Statistics and Computing, 17(4):395–416. find
  54. Michael L. Waskom. 2021. seaborn: statistical data visualization. Journal of Open Source Software, 6(60):3021. Publisher: The Open Journal. find
  55. Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pier-ric Cistac, Tim Rault, Remi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mariama Drame, Quentin Lho… — Transformers: State-of-the-Art Natural Language Processing

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/binkowski-2025-hallucination-detection-llms-spectral-features/